Comment 11 for bug 1927004

"Also note that while 'base' and 'extra' might have been used before, you might not have seen Seth's comment 7 that indicated it didn't turn out very well for apparmor, so maybe 'base' and 'extra' aren't the best names, even if they've been used before."

Thanks, I did see that and we discussed it. That was my direct reply about avoiding terms that imply judgment.

I appreciate that it would be good to have a common ground on how to go about this problem. That said, in our discussions, we're finding different cases at play. In our case, the agents that we're looking to place into Main are ones that we feel cover some 80% of the typical use cases along with some particular ones that our team has an interest in adding testing and validation.

The other agents are not bad or any less, it's just an admission that the ones in Main our team will be working on that testing, adding documentation to the server guide, etc. We cannot provide that for all of them and thus the split to help reality and expectations align.

Honestly, most use cases will only leverage a single fence-agent or resource-agent and in an ideal world they'd all be independent packages that could be treated as individuals but the burden of 45+ different packages for this would be unsustainable from the team.

In our case -core doesn't work because you don't have to have that package installed to use anything from -extras. You also can't replace using something in -extras with something in base. Good/bad are not legitimate distinguishers imo. Supported is indeed overloaded and not the best wording.

In the end, I don't know that a single "this is how to do it" fits in the myriad of use cases we've discussed so far. Given that there's not a current single-use case I've encouraged the team to not let perfect be the enemy of the good and help move forward the work we're doing to build out a well-supported set of tools for HA on Ubuntu.

I very much appreciate the feedback and think we're heading in a better direction than started. I'm happy to dive more into the discussions and if there's a better pattern that's been identified that works with the concerns I've raised I'm happy to learn about them.